R
Member
A patch antenna that measures fine on the bench often shifts badly once it goes into a metal housing. I ran into this on a small IoT module: the return loss looked great in free space, then the center frequency moved almost 200 MHz inside the enclosure.
First, check the spacing, not just the patch size. A ground plane or metal wall within a few millimeters loads the fringing fields and pulls the resonance down. As a rule of thumb, I keep at least a quarter-wavelength of clearance where the board allows it, and I re-measure at every spacing step instead of trusting one free-space tune.
Second, mind the ground plane size. A patch tuned on a large evaluation ground will behave differently on a small product ground, because the ground itself is part of the radiator. When the product ground is much smaller, I retune on a mockup of the real ground rather than the eval board.
Third, use a slot or a shorting pin deliberately. A small slot near the feed can compensate for the capacitive loading from nearby metal, and it is cheaper than respinning the patch geometry. But tune it after the housing is fitted, not before, or the compensation lands in the wrong place.
Fourth, watch the cable and connector in the test setup. A long coax pigtail near metal can add its own resonance and make the housing look worse than it is. I route the test cable away from the housing edge and repeat the sweep with the cable moved, just to confirm the dip is real.
Finally, log the detuning budget. Record the free-space frequency, the in-housing frequency, and the shift for each prototype, so the next housing revision starts from data instead of guesswork.
Have you had a patch that tuned perfectly on the bench but fell apart in the enclosure? What spacing or grounding fix actually recovered it?
First, check the spacing, not just the patch size. A ground plane or metal wall within a few millimeters loads the fringing fields and pulls the resonance down. As a rule of thumb, I keep at least a quarter-wavelength of clearance where the board allows it, and I re-measure at every spacing step instead of trusting one free-space tune.
Second, mind the ground plane size. A patch tuned on a large evaluation ground will behave differently on a small product ground, because the ground itself is part of the radiator. When the product ground is much smaller, I retune on a mockup of the real ground rather than the eval board.
Third, use a slot or a shorting pin deliberately. A small slot near the feed can compensate for the capacitive loading from nearby metal, and it is cheaper than respinning the patch geometry. But tune it after the housing is fitted, not before, or the compensation lands in the wrong place.
Fourth, watch the cable and connector in the test setup. A long coax pigtail near metal can add its own resonance and make the housing look worse than it is. I route the test cable away from the housing edge and repeat the sweep with the cable moved, just to confirm the dip is real.
Finally, log the detuning budget. Record the free-space frequency, the in-housing frequency, and the shift for each prototype, so the next housing revision starts from data instead of guesswork.
Have you had a patch that tuned perfectly on the bench but fell apart in the enclosure? What spacing or grounding fix actually recovered it?
